A statistically significant ethanol storage-loss equation is not necessarily accurate enough to correct an individual result. In a 2007 study of preserved blood, the investigator explicitly warned against using the regression to predict a specimen’s loss because individual variation remained large. Applying that equation to postmortem blood would add a further, untested transfer between specimen populations.
Three datasets served different purposes
Jones studied venous whole blood from living drivers in Sweden. Blood was collected from an antecubital or cubital vein into 10-mL evacuated glass tubes containing 100 mg sodium fluoride and 25 mg potassium oxalate. Tubes were mixed after collection and refrigerated at 4°C.
The study used 4,739 cases to characterize analytical imprecision across ethanol concentrations. A separate set of 159 specimens supported the storage-change analysis for intervals up to 12 months. Another experiment used paired tubes from 49 drivers to compare repeated opening with unopened storage. These denominators should not be combined into one storage cohort.
None of these groups consisted of decomposed autopsy specimens. The paper discussed microbial difficulties with postmortem blood, but did not experimentally validate its storage model in that matrix.
A precise assay still showed variable specimen change
Ethanol was determined in duplicate by headspace gas chromatography, with independent technicians and equipment. At the mean blood ethanol concentration of 1.64 mg/g, the estimated standard deviation of a single determination was 0.019 mg/g, corresponding to approximately 1.1% CV.
Among the 159 storage specimens, the mean decrease was 0.105 mg/g with SD 0.0686 mg/g. The change was statistically significant, p < 0.001. Loss correlated with storage duration, r = 0.44, and with starting concentration, r = 0.23. Both associations were statistically significant, but neither implied that a particular tube followed the average.
One specimen showed a small increase of 0.006 mg/g after two weeks, which was analytically insignificant. The study found no significant increase under its tested conditions. This is different from demonstrating that all preserved postmortem specimens are protected against ethanol formation.
The regression and its explicit limitation
The reported equation was:
Predicted decrease in mg/g = 0.0306 + 0.000306 × storage days + 0.0151 × the study’s blood ethanol term (BEC), in mg/g.
The multiple correlation coefficient was 0.467, with p < 0.001. Both predictor coefficients were statistically significant. However, the residual standard deviation was 0.061 mg/g. The author therefore stated that the equation should not be used to predict individual storage loss.
Statistical significance indicates that the observed association was unlikely to be explained by the tested null model. It does not make unexplained between-tube variation disappear. Adding a calculated “lost amount” back to a later result would present an individual correction the source expressly declined to support.
Opening history supplied another source of variation
In the 49-driver paired-tube experiment, tubes opened 17 times over 6.5 months lost an average of 0.217 mg/g, compared with 0.101 mg/g in tubes kept unopened. Their respective SDs were 0.054 and 0.076 mg/g; the difference was significant, p < 0.001.
Those group means do not provide fixed corrections for opened and unopened tubes. They show why storage duration alone does not describe the complete specimen history. Container integrity, handling and preservative conditions remain relevant. Our article on opened and underfilled tubes discusses related handling questions.
The additional limit for a death investigation
The results were reported in mg/g whole blood. They are not automatically interchangeable with serum concentrations or with a volume-based result expressed in g/L. More fundamentally, samples collected from living drivers under the stated conditions do not reproduce bacterial contamination, decomposition or precollection change in a cadaver.
The study can support asking for documented handling conditions and method imprecision. It cannot reconstruct ethanol at death, prove antemortem drinking or quantify postmortem formation. A separate serial-testing postmortem case illustrates why formation and storage loss are separate possibilities that require their own evidence.
Primary source
Jones AW. Are changes in blood-ethanol concentration during storage analytically significant? Importance of method imprecision. Clinical Chemistry and Laboratory Medicine. 2007;45:1299–1304. https://doi.org/10.1515/CCLM.2007.289.
Leave a Reply